Title: Network benefits of high hash power. Post by: pletharoe on October 15, 2013, 07:58:13 AM I'm having trouble understanding the benefits to the network of having so much computing power. It seems to be doing a lot of clever maths, most of which is wasted.
Below I have calculated the computing power required to process one transaction. Most of this is done on whizz bang ASIC machines, but how long would a single transaction take on one PC? Current network computing power: 22,974.66 Petaflops. Petaflop per day: (22974.66*60*60*24) = 1,985,010,624 Transactions in last 24 hours: 57,084 Petaflop per transaction: (1,985,010,624 / 57,084) = 34,773.5 An 8 core 2.5GHz computer can operate at around 80 Gigaflops. (Taken from http://en.wikipedia.org/wiki/FLOPS Single core is 10 Gigaflops, so 8 cores is 80 Gigaflops) Processing time required to complete a single transaction on a single 8 core machine: 34,773.5 Petaflop / 80 Gigaflops = 434,668,750 seconds. That's 13.78 YEARS! Is there any benefit in using so much computing power? Does it make the network more secure? Couldn't we use this power to reduce the confirmation time? Maybe by producing a block every minute (and adjusting the BTC reward to miners accordingly). This would make BTC a much more viable every day currency. Title: Re: Network benefits of high hash power. Post by: OnkelPaul on October 15, 2013, 08:08:13 AM That's a standard newbie question.
Short answer: the goal of the bitcoin block hashing mechanism is to make the system secure. To achieve that goal, a "proof of work" mechanism is used which ensures that to break the security more than half of the total hashing power needs to be controlled by the bad guy. This goal contradicts directly with the efficiency goal: If the algorithm were designed to use only little resources, its security could be broken with only little resources - i.e. the system would be totally insecure. Maybe this comparison helps to understand the issue: You could cross a river by just throwing a rope to the other side and fixing it there. This would be far cheaper and efficient than building a bridge using thousands of tons of concrete and steel. However, your rope bridge would not be suitable for vehicle traffic carrying goods, it could only be used by artists balancing on the rope and would immediately snap if you tried to use it for transporting a ton of goods. Onkel Paul Title: Re: Network benefits of high hash power. Post by: enquirer on October 15, 2013, 08:17:05 AM Instead of making 5 useless posts, newbies should be forced to read and type "I understand" under 5 FAQ questions, one of which is above.
Title: Re: Network benefits of high hash power. Post by: pletharoe on October 15, 2013, 08:31:17 AM Thanks OnkelPaul - Your input was very helpful.
No thanks enquirer - You have proved the old adage that there is no such thing as a stupid question, only a stupid answer. I was already aware of the danger posed by a single party gaining control over too much of the network. However, it still seems to me that more frequent blocks would allow for quicker transaction confirmations and would still allow the same level of security. Title: Re: Network benefits of high hash power. Post by: OnkelPaul on October 15, 2013, 08:41:58 AM I was already aware of the danger posed by a single party gaining control over too much of the network. However, it still seems to me that more frequent blocks would allow for quicker transaction confirmations and would still allow the same level of security. I think that's a FAQ, too... Since block generation is a poisson process, the actual time between two blocks can be much less than the average time. If the time between two blocks is less than or close to the full network propagation time, the probability of orphan blocks rises significantly. Orphans essentially weaken security because when two versions of the blockchain are considered valid by different parts of the network, and the conflict is not resolved quickly, there is a risk of involuntary double spending / reversed transactions. Onkel Paul Title: Re: Network benefits of high hash power. Post by: b!z on October 15, 2013, 10:24:07 AM From what I understand, the faster the network is hashing, the harder it would be for an attacker to take control of the network, through a 51% attack or something similar.
Title: Re: Network benefits of high hash power. Post by: pletharoe on October 15, 2013, 02:01:55 PM Thanks again OnkelPaul. I hadn't accounted for the time it takes for a confirmation to propagate through the network.
Title: Re: Network benefits of high hash power. Post by: hayek on October 15, 2013, 03:44:30 PM Instead of making 5 useless posts, newbies should be forced to read and type "I understand" under 5 FAQ questions, one of which is above. Or better yet, re-write 5 FAQ questions in their own words using real world examples Title: Re: Network benefits of high hash power. Post by: pletharoe on October 16, 2013, 05:09:03 AM Instead of making 5 useless posts, newbies should be forced to read and type "I understand" under 5 FAQ questions, one of which is above. Or better yet, re-write 5 FAQ questions in their own words using real world examples Or even better, get a PhD in cryptocurrencies. Title: Re: Network benefits of high hash power. Post by: davidgdg on October 16, 2013, 09:24:25 AM That's a standard newbie question. Short answer: the goal of the bitcoin block hashing mechanism is to make the system secure. To achieve that goal, a "proof of work" mechanism is used which ensures that to break the security more than half of the total hashing power needs to be controlled by the bad guy. This goal contradicts directly with the efficiency goal: If the algorithm were designed to use only little resources, its security could be broken with only little resources - i.e. the system would be totally insecure. Maybe this comparison helps to understand the issue: You could cross a river by just throwing a rope to the other side and fixing it there. This would be far cheaper and efficient than building a bridge using thousands of tons of concrete and steel. However, your rope bridge would not be suitable for vehicle traffic carrying goods, it could only be used by artists balancing on the rope and would immediately snap if you tried to use it for transporting a ton of goods. Onkel Paul But of course any engineer has to identify the correct balance between strength and cost. You wouldn't use thousands of tons of concrete and steel to ford a creek. I think what lies behind the OP's question is really whether the current balance has gone too far in favour of security at the expense of speed of confirmation. That is a fair question, though the answer may not be straightforward. Title: Re: Network benefits of high hash power. Post by: hope4me on October 16, 2013, 09:41:26 AM But of course any engineer has to identify the correct balance between strength and cost. You wouldn't use thousands of tons of concrete and steel to ford a creek. I think what lies behind the OP's question is really whether the current balance has gone too far in favour of security at the expense of speed of confirmation. That is a fair question, though the answer may not be straightforward. With faster blocks like Litecoin, you see higher Orphan rate when the block contains alot of transactions (not Litecoin case :P), thus higher Orphan rate = wasted resources Title: Re: Network benefits of high hash power. Post by: OnkelPaul on October 16, 2013, 10:58:26 AM of course any engineer has to identify the correct balance between strength and cost. You wouldn't use thousands of tons of concrete and steel to ford a creek. I think what lies behind the OP's question is really whether the current balance has gone too far in favour of security at the expense of speed of confirmation. That is a fair question, though the answer may not be straightforward. That's right, and my example was probably a bit exaggerated. However, Bitcoin is being used to moves values to the tune of several million dollars around - I'd rather like it as secure as possible at the expense of some speed... Onkel Paul |